A publicly-available MD COVID-19 metrics that I didn’t investigate much is cases per ZIP code. I created a dashboard where you can highlight one zip code at a time. Tyler Fogarty built a cool Treemap Explorer. Silver Chips has a nice heatmap of all zip codes as part of their extensive dashboard (a bit like the MDH dashboard). How can we make sense of all this?
Today (May 30, 2020), 4 ZIP codes have more than 1,000 cumulative cases: 2 in Prince George’s county (20783, Hyattsville, and 20706, Lanham) and 2 in Montgomery county (20906 and 20902, both in Silver Spring). But among the ZIP codes with the most recent daily positive cases, 21223 and 21224 are also in the top 5, both in Baltimore City. All these ZIP codes are in counties that are closed or partially opened, highlighting the need for these regions to remain vigilant and enforce stay at home, wearing a mask and social distancing (at least).
I’m looking at protests in Baltimore and I can’t imagine how detrimental the spread of coronavirus will be and will add to the other issues. Here is what Prof. Murray advise to protect protested (on top of any other precautions):
There is no public data about hospitalization per ZIP code nor deaths per ZIP code. But there are certainly other ways to make sense of this metric …
Since my previous post (May 21), tests were broadened in some drive-thru locations for anyone to be tested (5/19 actually) and new testing sites were opened (map of sites here) and we had the Memorial Day weekend (5/25). On May 28, Gov. Hogan mentioned that “hospitalizations, ICUs, and testing positivity rates are the key metrics in determining Maryland’s road to recovery”. On May 27, Gov. Hogan announced that further reopening were taking place (outdoor dining, some outdoor activities for kids allowed, …) but still within Stage 1 (I called it “Stage 1b”).
In terms of hospitalizations, the graph above shows the number of patients currently hospitalized (green line). Since beginning of May, hospitalizations decreased, especially thanks to the decrease of patients in acute care (red line). Patients in ICU (Intensive Care Unit) decreased much slowly. This is probably due the severity of these patients, making them stay for a long time and released from ICU at a slower rate than patients in other departments. And the graph also shows that Maryland never needed the additional hospital beds prepared for a worst scenario.
The third key metric the Governor is looking at is testing positivity rate. The chart above represents, on top, the total number of tests reported on the MD Health Department dashboard (adding positive and negative test results). We learned that the Governor is actually not looking at the same positivity rate than the one we can compute from the dashboard:
The dashboard report unique positive and negative tests. If someone was tested twice or more with the same result, it would have been reported only once. If the test result would change, it would have been reported once in each category.
The Governor is looking at all positive and negative tests. If someone was tested twice or more, independently of the result, all tests results would have been counted here.
This difference probably explain why we see a daily number lower than 10,000, despite 500,000+ tests received by Governor from South Korea. But we can’t really say in which direction this difference would drive the testing positivity rate. If more positive tests were under-counted (i.e. counted once instead of the several times they were performed/received), the Governor would have seen a higher positivity rate than on the dashboard. More likely, if more negative tests were under-counted (i.e. negative people tested several times, but counted once), the Governor would have seen a lower positivity rate than on the dashboard. This last option would explain why the Governor decided to go on Stage 1 sooner than expected by just watching the dashboard.
Technically, as a side note, the data for the testing positivity rate that the Governor is looking at is not publicly shared. There is just a PDF with graphs. This difference in what is reported may also explain why, since test broadening (5/19), there was 5 days of ups and downs after which the rate stayed at about 10%.
At the level of the State of Maryland, we are not yet looking at the full picture: the last element (that doesn’t seem to be part of the key metrics) is deaths. So far (since mid-March), there have been 2,390 deaths due to COVID-19 in Maryland with a majority of them occurring in congregated facilities (nursing homes, prisons, etc.). With an about-weekly pattern (see below), the daily number of confirmed deaths also seem to decrease (although much slower than hospitalizations or positivity rate).
But if things seem good at the State level, the decision to reopen Maryland came with the empowerment of Counties (the government level below State) to follow or not the reopening. As noted before, if most counties followed the State in Stage 1, some counties did not (some like Prince Georges and Montgomery even remained “closed”). There is no straightforward way to follow hospitalizations in counties (they are not reported on the MD dashboard). But we can follow deaths in counties in the graph below. There it is a bit surprising to see that counties that re-opened, the % of deaths compared to May 15 is actually increasing (i.e. more daily deaths in counties that re-opened) (see blue dots and average in the blue line). On the other hand, % of daily deaths seems to decrease in counties that partially reopened or remained closed. But one should also note the huge confidence intervals around these averages.
Finally, about counties, the situation is about to get messier: since yesterday, Anne Arundel, Baltimore City and Howard counties further allowed some outdoor activities; and starting June 1st, Montgomery and Prince George’s counties, initially closed, will also start to allow some outdoor activities.
Here is a first attempt to look at the fate of the different counties. My idea here is to set the number of cases in all counties on May 14, 2020 (start date of Stage 1) to 100% and see how counties evolve in terms of number of new daily cases.
On top of the figure below, I represent the cumulative, 7-day average (*) daily new COVID-19-confirmed cases in the different counties of Maryland. The chart at the bottom assign the number of daily cases on May 14, 2020 to 100% for each state and follow the % evolution over the next day. In this chart, the blue lines represents counties that follow Stage 1 (e.g. Garrett or Ken), the green line represents counties that partially follow Stage 1 (e.g. Anne Arundel or Frederick) and the red line represents counties that remain “closed” (Baltimore City, Charles, Prince Georges and Montgomery). The counties that remain closed are the ones that have the most cases and deaths.
I must say that 6 days after Stage 1 (May 20), there is no clear trend. First, it’s normal because any downward or upward trend in number of cases will take a few day to appear (transmission or absence of transmission, incubation, decision to consult and tests, and lag in test reporting). It’s too early to see something. We will also see a confounding factor with the recent decision by the Governor to allow testing of people who do not present any symptoms (in some testing sites). Nevertheless, I was expecting to “see something”; here it just seems it’s the same.
But another reason for “not seeing anything” might be that the cases are not a relevant metric. We can already see that it is fluctuating widely every day. There are even days when less cases were reported than the day before (it might have been a data entry error on my side). The only other parameter that the MDH displays in its dashboard is the number of deaths by counties. I plotted this and it’s the same bizarre chart. How to improve this? Any idea? Don’t hesitate to comment below or to send me an email.
Update on May 24, 2020: I updated the chart of cases after Stage 1 (see above). Currently the confidence intervals (the shades) are so overlapping that differences that we could see are meaningless. Cases may not be a good metric.
I also created the same chart for deaths (see below). Here we see clearly a positive picture: in all counties that are partially open or closed, the mean number of deaths is decreasing. Note however that we are only 10 days (today is 5/24) from May 15 and this may just be a trend that existed before and not something new due to the decision to remain (partially) closed.
In counties that are in Stage 1, the mean number of deaths is actually increasing. The same comment applies: it may be too early to actually see an impact of the opening (especially deaths could be far from the case detection). Besides, the confidence intervals (the blue shades) are very wide. Hopefully things may become clearer in a few days (and for the best, given we are talking about a disease and people dying from it).
Regarding testing … After a peak in testing (up to 8k/day) and the arrival of tests from South Korea, testing is stagnating ~ 4k/day in Maryland. The daily % of positive tests seem to stagnate ~ 22% since May started and decline a bit, ok. Despite a peak, today, we are still far from the daily number of tests we could reach with the tests from South Korea. And there is still a high percentage of daily positive cases.
Regarding hospitalizations … Thanks to a drop in acute care during the last weekend and this one, the number of patients hospitalized seems to decrease since beginning of May. This sudden drop was followed by a slight re-increase. Hopefully this will continue to decrease (even if in waves like this).
Regarding deaths … Despite a record number of daily reported deaths, early May, and wide variations in this daily metric, it seems that we are plateauing/decreasing here.
So, we can cautiously understand the decision taken, based on data. Note it’s not a total opening: the “flexible community-based approach” means that counties (the government level below the state of Maryland) are empowered to make decisions regarding actually opening or not. And some counties took that opportunity. For instance, Montgomery, Prince George’s and Charles (on the East / South-East of Maryland, surrounding Washington DC and bordering Virginia) decided to remain closed. Baltimore City also decided to remain closed. It is understandable as all these counties are among the ones with most cases and most deaths. Here is a small infographics summarizing the Stage 1 reopening:
It will be interesting to see how the next 2 weeks evolve, and especially if counties that remain closed will have a different evolution than the ones that opened.
Since a few weeks, I report the raw number of COVID-19 deaths in Maryland counties. If this gives an idea of the cumulative number of deaths – which is interesting – it doesn’t reflect the fact that some counties have more inhabitants than others. That’s why I plotted below the number of COVID-19 deaths adjusted for the population (i.e. the COVID-19-specific death rate):
Today (May 16, 2020), in terms of absolute number of deaths, Montgomery, Prince Georges and Baltimore County are the top 3 counties (this is the same for cases but not in the same order). In terms of confirmed deaths per 100,000 population, the top 3 counties are Kent, Prince Georges and Montgomery.
Since a few weeks, I report the raw number of COVID-19 cases in Maryland counties. If this gives an idea of the cumulative number of cases – which is interesting – it doesn’t reflect the fact that some counties have more inhabitants than others. That’s why I plotted below the number of COVID-19 cases adjusted for the population:
Today (May 11, 2020), in terms of absolute number of cases, Prince Georges, Montgomery and Baltimore County are the top 3 counties. In terms of confirmed cases per 100,000 population, the top 3 counties are Prince Georges, Montgomery and Wicomico (due to a recent surge in cases).
Rank on May 11, 2020
Absolute # of COVID-19 cases
COVID-19 cases per 100,000 population
Prince Georges (9,687)
Prince Georges (1,057)
Baltimore County (3,948)
Baltimore City (3353)
Anne Arundel (2492)
Baltimore City (544)
This is a lot given that, today, the average for Maryland is 401/100,000 (source: CDC) and the average for the US is 552/100,000 (source: OurWorldInData).